gnodet commented on code in PR #25899:
URL: https://github.com/apache/camel/pull/25899#discussion_r3889142674
##########
components/camel-ai/camel-huggingface/src/main/java/org/apache/camel/component/huggingface/tasks/AbstractTaskPredictor.java:
##########
@@ -107,6 +107,22 @@ public void setEndpoint(HuggingFaceEndpoint endpoint) {
protected abstract String getPythonScript();
+ /**
+ * Prepends the configured Hugging Face token to the generated handler as
the {@code HF_TOKEN} environment variable
+ * so that every task can load gated or private models. {@code
transformers.pipeline()} reads {@code HF_TOKEN} from
+ * the environment when no explicit token is passed; previously only the
chat task passed a token, so the other
+ * tasks failed with 401 on gated models. The token comes from the {@code
authToken} option or is resolved from an
+ * OAuth profile (see {@link
org.apache.camel.component.huggingface.HuggingFaceProducer}), both surfaced
through
+ * {@code config.getAuthToken()}.
+ */
+ protected String withAuthToken(String pythonScript) {
+ String authToken = config.getAuthToken();
+ if (authToken == null || authToken.isEmpty()) {
+ return pythonScript;
+ }
Review Comment:
Now that the token is applied centrally here via `HF_TOKEN`, the
`tokenClause` in `ChatPredictor.getPythonScript()` (line 113) is redundant —
chat gets the token through both the env var and the explicit `token='...'`
kwarg. Consider removing the `tokenClause` logic from `ChatPredictor` and the
`%s` placeholder from `chat.py` to match the PR's intent of centralizing token
handling.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
